In a prospective, randomised trial, we compared the use of three Ullevaal hip screws with that of two Hansson hook-pins in 278 patients with fractures of the femoral neck. Background factors were similar in both groups. Follow-up was for two years. There were no significant differences between the groups in length of time of surgery, hospital stay, general complications, mortality, pain or walking ability. Likewise, the rates of early failure of fixation, nonunion, and the need for reoperation did not differ significantly between the groups. The use of hook-pins was associated with less drill penetrations of the femoral head during surgery (odds ratio 2.6, p= 0.05) and a lower incidence of necrosis of the femoral head (odds ratio 3.5, p = 0.04). There was a strong relationship between poor reduction and fixation of the fracture and subsequent reoperation (p = 0.0005 and p = 0.0001, respectively). Likewise, peroperative drill penetration of the femoral head was associated with a greater risk of reoperation (p = 0.038). Both methods gave favourable results. In total, 22% of the patients needed a major reoperation (usually hemiarthroplasty), while in 7% of the cases the fixation device needed to be removed. Osteosynthesis as the sole method for operation of all fractures of the femoral neck was thus successful in 78% of patients. With selective treatment most of the remaining patients would have benefited if treated by a primary arthroplasty. Accurate selection requires the development of better prognostic methods